A Statistical Evaluation of Asbestos Air Concentrations

Abstract

Both area and personal air samples collected during an asbestos abatement project were matched and statistically analysed. Among the many parameters studied were fibre concentrations and their variability. Mean values for area and personal samples were 0.005 and 0.024 f·cm^–3 of air, respectively. Summary values for area and personal samples suggest that exposures are low with no single exposure value exceeding the current OSHA TWA value of 0.1 f·cm^–3 of air. Within- and between-worker analysis suggests that these data are homogeneous. Comparison of within- and between-worker values suggests that the exposure source and variability for abatement are more related to the process than individual practices. This supports the importance of control measures for abatement. Study results also suggest that area and personal samples are not statistically related, that is, there is no association observed for these two sampling methods when data are analysed by correlation or regression analysis. Personal samples were statistically higher in concentration than area samples. Area sampling cannot be used as a surrogate exposure index for asbestos abatement workers.

Statistical Analysis of Exposure Data: Formaldehyde in Occupied and Unoccupied Caravans in Australia

“…It has been suggested that summary exposure data be reported as arithmetic mean (AM), geometric mean (GM), standard deviation (SD), geometric standard deviation (GSD) and range [2][3][4]. In the aforementioned article, some of these descriptors were provided.…”

“…It has been generally accepted that environmental (area) and occupational (personal) exposure concentration distributions are nonnormal with most having a logarithmic form [2][3][4][7][8][9][10]. Previous investigations of formaldehyde exposure [10][11][12] suggest that concentrations of this chemical (exposure) follow a nonnormal distribution, which is most likely lognormal.…”

“…Based on previous experience, it is likely that the Pearson's product-moment correlation test may be appropriate for duplicate samples and the Spearman rankorder correlation test for occupied and unoccupied caravans. Another important consideration in analysis of exposure data is determination of outliers [3,15]. Identification of extreme values may prove useful in interpretation of data.…”

“…arcsine, square root, log) if there is no previous information on its form [6,16]. However, for exposure data, it is suggested that the W test is most appropriate, particularly for sample populations with less than 50 measurements [2][3][4]. For exposure samples, including area samples as reported in the article [1], previous studies [3,7,10,11,17] have suggested the distribution to be nonnormal and to best fit a logarithmic function.…”

“…When transforming data for application of parametric methods, the appropriate method of transformation must be employed [16], thus requiring actual knowledge of the distribution. From this scenario, it is suggested that nonparametric methods be employed for most exposure analyses [2][3][4][19][20][21][22][23]. Regardless, methodologies of statistical analysis must be described [3] so readers can fully evaluate methods employed for their own interpretation of presented data.…”